The Ten Commandments of Therapeutics

I found a wonderful article by Udkin, Dreger and Sousa, and have updated and adapted it. See below the fold…

The New Therapeutics: Ten Commandments

(Adapted from work by Dr. John Yudkin, Dr Alice Dreger and Dr Aron Sousa.)

1. Thou shalt treat according to the patient’s level of risk rather than level of risk factor. A risk factor is a trait that increases the likelihood of having something really bad happen to you. For example, having a high LDL is a risk factor for having a heart attack.  However, lowering your LDL level with some prescription drugs won’t actually decrease your risk of a heart attack. It is important to treat the patient, not just what is measurable.
2. Thou shalt exercise extreme caution when adding drugs to existing polypharmacy.  It is risky to give a patient another drug when a patient is already on a drug. It increases the risk of interactions between the drugs.  There is roughly a 15% chance of a drug interaction when adding a second drug, 40% when adding a 5th drug and 85% when adding an 8th drug. Adding drugs also increases the chance of medication error. 
3. Thou shalt consider benefits of drugs as proven only by hard endpoint studies.  What you really care about are “hard endpoints” –  heart attacks and strokes, not levels of cholesterol , BP or A1c. A drug might make lab results look really great, but it’s not a good drug if it doesn’t actually improve health in the ways that matter.  A drug might increase bone density or lower blood pressure or cholesterol, but if it doesn’t reduce your risk of fracture, stroke or heart attack, who cares that your bones are denser, your BP is lower or cholesterol is better?
4. Thou shalt not bow down to surrogate endpoints, for these are but graven images.  Surrogate endpoints are things like blood pressure readings, as opposed to ‘hard endpoints’ like heart attacks and strokes. It doesn’t actually matter what your blood pressure is if your blood pressure doesn’t hurt you. We use blood pressure readings as a surrogate for what we really care about. It’s worth measuring, but we should not treat high blood pressure with a drug unless that particular drug is shown to achieve what we really care about: reduction of risk of heart attacks and strokes.
5. Thou shalt not worship Treatment Targets, for these are but the creations of Committees.   Consensus groups come up with ‘treatment targets’  to tell physicians what an average patient's lab numbers should be. Average patients are truly rare. Clinicians should treat individual unique patients, not average patients or populations. An LDL of 140 should not be treated the same way in the smoking, hypertensive, obese diabetic with a strong family history of heart attacks as in the 35 year old non-smoking elite endurance athlete with a BP of 106/60 whose family members die in their 90s participating in extreme sports. Physicians and patients should especially beware any consensus issued by committees of people who have financial ties to drug and device makers.”
6. Thou shalt apply a pinch of salt to the abomination of Relative Risk Reductions, regardless of P values, for they are a marketing ploy designed to mislead.  This is a complicated way of reminding physicians what should matter: actual reduction of the risk of things their own real patients care about and are likely to suffer from. Relative Risk Reduction is another way drug companies often fool physicians and patients into thinking a drug is better than it really is.
7. Thou shalt honour the Numbers Needed to Treat (NNT), for therein rest the clues to patient-relevant information and to treatment costs.  ‘Numbers needed to treat’ refers to how many patients a doctor needs to treat with a particular intervention in order to have one patient benefit. Not every intervention benefits every patient. There’s a ‘number needed to harm’ for each drug, too, but you don’t often hear about either NNT or NNH from your doctor. For most medical interventions, you must treat a relatively large number of people to benefit just one while exposing ALL of them to the risks of that intervention. 
8. Thou shalt use natural frequencies rather than probabilities and percentages, and use the smallest numbers possible.  The statement ‘8 out of 10 patients like you will benefit from this treatment’  is more easily understood than 80% of patients benefit or your chance of benefiting is 80%. Telling a patient that  12,685 out of 15,211 benefit is confusing and saying 83.4% implies a precision not justified by the data. 
9. Thou shalt share decisions on treatment options with the patient using estimates of the patient’s individual potential for risks and benefits.  The patient and clinician should first decide what the patient is trying to achieve, and then discuss the options and evidence. 
10. Honour the elderly patient, for although this is where the greatest levels of illness reside, so do the greatest hazards of many treatments.   Self-explanatory, but too often ignored.